The Impact of Human Milk on Outcomes for Infants with Congenital Heart Disease: A Systematic Review

Operational definitions

For the purposes of this review, operational definitions are as follows: (1) “Infants” are children <1 year of age. (2) “Congenital heart disease” is determined by confirmed clinical diagnosis. (3) “Human milk feeding” will be defined per study, and an examination of how this is defined is a secondary question of interest. (4) “Other types of feeding” will be defined per study and may include formula feeding, bovine-based fortification, or nothing by mouth (npo). (5) “Outcomes” will not be defined a priori, but will be defined per study.

Due to high variability in terminology, “human milk” will be used as an umbrella term to indicate mother's own milk (MOM) and/or donor HM, with the recognition that these two nutritional forms are not interchangeable. ²³ An investigation of differences associated with the route of nutrition (e.g., direct latch at the breast, bottle, nasogastric tube) is beyond the scope of this review.

Search strategy

Following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines, ²⁴ a literature search was conducted with the assistance of a university-based health sciences librarian, using MEDLINE (accessed via Ovid MEDLINE^® and Epub Ahead of Print, In-Process & Other Non-Indexed Citations, 1946 to 2021), CINAHL, and Cochrane Database of Systematic Reviews. Key subject headings included “human milk,” “breast feeding,” “heart defects, congenital,” “cardiac surgical procedures,” “infant,” and “infant, newborn.” Related keywords were added with truncation for maximum results. The search was limited to human subjects and English-language results, and review articles were excluded. The full Ovid MEDLINE search strategy can be seen in Table 1.

Study selection

Articles were imported into Rayyan QCRI ²⁵ and duplicates were removed. An initial blinded screen by title and abstract was completed by two reviewers (K.M.E. and A.C.M.). Studies that appeared to fit the inclusion criteria and those with abstracts without enough information to determine eligibility received a blinded full-text review by the same two reviewers. Reference lists of included studies were examined for additional eligible publications. Any disagreements were solved through consensus.

Articles were included in the review if (1) the study population was infants with clinically diagnosed CHD; (2) the impact of HM on outcomes for infants with CHD was a primary or secondary focus; and (3) the article reported findings from an original research study, including observational, quasiexperimental, or experimental designs. Conference abstracts were considered if there was enough information available to make a quality assessment.

Articles were excluded if (1) the study population was focused on infants with a primary diagnosis other than CHD (e.g., a genetic syndrome); (2) the article was written in a language other than English; (3) the article was not original research (e.g., review, expert opinion); and (4) the article reported findings from a case study or a quality improvement project. The search resulted in 483 records. After deduplication, title and abstract review, and full-text examination, 16 articles were included in the review. Figure 1 is a PRISMA flow diagram of the search process.

OPEN IN VIEWER

FIG. 1.

PRISMA flow diagram of the search process. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-analyses.

Data extraction and quality assessment

Data from the included studies were extracted by one author (K.M.E.) and validated by a second author (A.L.T. or K.M.S.). Information collected included citation, study methods, sample characteristics, results, limitations, and funding sources. First authors were contacted for missing data.

The methodological quality of each study was assessed using the Critical Appraisal Tools from the Joanna Briggs Institute. ²⁶ These are a collection of reliable, validated, design-specific tools that can be used to evaluate the quality of observational, quasiexperimental, and experimental studies. An initial quality assessment was conducted by the first author (K.M.E.), with a second blinded assessment by an additional author (A.C.M., E.N.S., N.M.S., or K.M.S.). Any disagreements were solved through consensus.

Study characteristics

A total of 16 articles met the final inclusion criteria. Nine were cohort studies (seven retrospective^27–33 and two prospective^34,35), four featured a quasiexperimental design,^36–39 two were randomized-controlled trials (RCTs),^40,41 and one was a retrospective case/control design. ¹⁹ One RCT included multiple sites, ⁴⁰ and two studies were secondary analyses of multisite data sets.^19,32 The studies took place most often in the United States (n = 10), with Canada (n = 3), China (n = 1), Germany (n = 1), and Italy (n = 1) also being represented. A summary of characteristics of the included studies can be found in Table 2.

Sample

There were a total of 8,176 participants in the 16 included studies.* These participants were newborns diagnosed with complex CHD,^27,29,32 medically stable newborns with CHD, ³⁴ infants undergoing cardiac surgery for CHD,^{19,30,31,33,35,38,40,41} or infants with CHD diagnosed with chylothorax after surgery.^28,36,37,39 Participants were 60.3% male and 39.7% female (female sex was often presumed, as only six studies reported female numbers). Only seven studies reported details of race and ethnicity.^{19,27–29,32,33,35} Of 7,659 infants, 56.5% identified as white, 26.5% as Hispanic, 11.7% as black, 0.3% as Asian, and 5.0% as other. Social determinants of health were reported in three studies, with Kocel et al. ³⁶ and Siemienski et al. ³⁵ detailing parental level of education, and McCrary et al. ³³ reporting the percentage of patients receiving Medicaid.

The age of the participants was most often reported at three time points, depending on the study aim. Gestational age, age at surgery, and/or age at chylothorax diagnosis. The mean or median gestational age was >37 weeks in 11 studies,^{19,27–30,33,36,37,39–41} >36 weeks in another, ³² and unreported in 4 studies.^31,34,35,38 Age at surgery was mean or median <3 months for all five studies reporting this metric.^{28,31,33,38,41} Age at chylothorax diagnosis varied, with four studies reporting a median range of 0.6–5.3 months.^28,31,36,37

Weight was also reported at the same three time points. The mean or median birth weight was >3.0 kg in seven studies,^{19,28–30,36,39,41} while Rosti et al. reported a birth weight range of 2.1–4.1 kg. ³⁰ Two other studies reported birth weights as percentages, with 77.4%> 2.5 kg in Becker et al. ³² and 93.4%> 2.0 kg in Cognata et al. ²⁷ Blanco et al. reported that 86.9% of infants were average or large for gestational age, ⁴⁰ and the remaining five studies did not report birth weight.^{31,33–35,38} Mean weight at surgery ranged from 2.6 to 4.0 kg in the three studies that provided these data,^28,30,38 while McCrary et al. reported a mean weight-for-age z-score (WAZ) at initial shunt placement of −0.61. ³³ Mean weight at chylothorax diagnosis varied, from 3.2 to 5.9 kg.^36,37,39

Three studies provide information about length and head circumference at birth and/or chylothorax diagnosis.^36,37,39 The median length at birth ranged from 47.5 to 51.0 cm,^37,39 while the median length at diagnosis was between 51.0 and 61.1 cm.^36,37,39 The mean or median head circumference at diagnosis ranged from 34 to 39.3 cm.^36,37,39

CHD diagnoses included both single ventricle (SV; n, % 3,291, 40.3) and biventricular (BiV; 4,885, 59.7) defects. Transposition of the great arteries was the most common BiV diagnosis, and hypoplastic left heart syndrome (HLHS) the most common SV physiology. Information about the diagnoses of the sample can be found in Figure 2.

OPEN IN VIEWER

FIG. 2.

Primary cardiac diagnoses of participants in the included studies.

Exclusion criteria varied among studies. Eight studies excluded infants with noncardiac congenital anomalies or genetic syndromes,^{27,29,30,35,38–41} and two studies excluded premature infants born <32 weeks.^33,37 Three studies excluded infants with noncomplex CHD (e.g., not requiring intervention or prostaglandin E1 infusion),^27,29,32 while Yu et al. excluded infants with complex CHD requiring staged surgery. ³⁸ Six studies excluded infants who experienced a complicated clinical course or death (e.g., preoperative necrotizing enterocolitis [NEC], chylothorax, extracorporeal membrane oxygenation, hemodynamic instability, or no enteral feeding).^{29,30,36,38,40,41}

HM feeding definitions

In six studies, infants in HM feeding groups received 100% MOM or donor HM.^{27,29,30,32,38,40} In one of these, infants in the HM feeding group received HM-based fortifier (Prolacta), while the comparison group received bovine-based fortifier. ⁴⁰ Two additional studies implied that HM feeding groups received 100% HM, but were unclear regarding MOM, donor HM, and/or fortification.^19,41 Four studies on treatment of chylothorax defined HM feeding groups as receiving low-fat MOM or donor HM with some type of fortification, such as medium-chain triglyceride (MCT) oil, MCT formula, soybean oil, or bovine-based HM fortifier.^28,36,37,39 Three of these studies considered varying amounts of formula supplementation acceptable in an HM diet, with ≥50%, ³⁹ ≥75%, ²⁸ or ≥80% ³⁶ required for HM feeding group eligibility.

Siemienski et al. defined a “breast milk predominant” group as receiving primarily HM (defined as MOM or donor HM) throughout the first year of life. ³⁵ In the final three studies, HM feeding was not clearly defined and allowed for any amount of HM, which led to a lack of differentiation between groups.^31,33,34 For example, in Boctor et al., 6 out of 10 infants in a “breastfed” group were fed MOM supplemented with formula, with no information available about HM dose. ³¹ Combs and Marino defined “breastfed” as any amount of MOM, ³⁴ while McCrary et al. compared infants with any “breast milk use” (with MOM or donor HM unspecified) with those with none. ³³

Findings

Outcomes examined in the studies can be grouped into four main categories: (1) incidence of NEC, (2) outcomes related to treatment of chylothorax, (3) weight gain, and (4) other postoperative outcomes. All significant results with point estimates can be found in Table 3.

Quality of evidence

The quality of evidence of each of the included articles was appraised using the Joanna Briggs Critical Appraisal Tools, ²⁶ with results presented in Table 4. It should be noted that quality assessment was focused solely on the strength of evidence as related to HM feeding and outcomes for infants with CHD, which was not a primary focus of all studies. The assessment does not necessarily reflect the strength of studies in regard to other outcomes.

As a group, the HM findings from the 16 studies exhibited methodological issues that call into question the validity of the results. Only four studies described clear, reliable outcomes measures (e.g., diagnosis of NEC and methods to track weight or other anthropomorphic measurements).^27,29,36,39

Seven of the nine cohort studies were retrospective chart reviews,^27–33 with inherent bias in data collection due to potential differences in accuracy between electronic health record users and potential for missing data. Most of these cohort studies did not report on baseline differences between feeding groups, with only Rosti et al. providing satisfactory evidence that the feeding groups were similar at study initiation. ³⁰ Of the four quasiexperimental studies, two did not confirm clear differentiation between the feeding groups, with some infants in HM groups consuming an appreciable amount of formula.^36,39

All but four studies^19,27,29,39 exhibited risk for bias regarding sample size, power, or statistical analysis, although this is understandable considering the small population and rare outcomes under study (e.g., NEC). Ten studies had very small samples or low incidence of outcomes (e.g., NEC), and/or did not provide information on whether power was adequate for HM-related outcomes.^{28,30–37,41} Five studies controlled for covariates,^{27,28,30,39,40} but it was unclear whether four of these were powered for all outcomes.^28,29,39,40 For example, Blanco et al. reported sufficient power for the primary outcome of weight velocity, but were unclear as to the power for NEC outcomes. ⁴⁰

Yu et al. completed a power analysis but did not control for covariates, and used potentially inappropriate tests for categorical variables with low frequency. ³⁸ Becker et al., Boctor et al., and Combs and Marino reported large amounts of missing data ³² or lacked adequate description of analytical methods,^31,34 which were significant limitations for results related to HM feeding.

Chylothorax

Chylothorax is another rare but serious complication of cardiac surgery, and treatment with a modified HM diet is a relatively new approach, with the first case study on this process published in 2004. ⁴⁷ In the context of care for chylothorax, outcomes were generally not different between modified HM groups and infants receiving MCT formula. While Neumann et al. reported significantly fewer treatment days for infants in a modified HM group, their results are skewed by one infant who received an MCT formula diet for 230 days. ³⁷ Considering the small sample size of the MCT group (n = 10), the true difference between groups is likely much lower. Differences in other treatment outcomes (e.g., number of chest tube days, volume of chest tube drainage, and LOS) were not significant.

The impact of a modified HM diet on growth in the context of chylothorax was inconclusive. Kocel et al. ³⁶ reported significant decreases in WAZ and LAZ from baseline, but included only 16 infants with a lack of definition between feeding groups. For example, infants in the modified HM group received as little as 45% modified HM, with the remainder supplemented by MCT formula. ³⁶ Conversely, 3 (38%) infants in the MCT group required parenteral nutrition between 5 and 19 days, compared with none in the modified HM group. ³⁶ This variation in feeding practice highlights the difficulty of conducting feeding research in infants with CHD—particularly when examining a rare outcome, such as chylothorax.

The remaining three studies found no significant differences in growth outcomes, although DiLauro et al. caution that they did not recruit the most severely ill infants, who may have been at highest risk for growth failure. ³⁹ In addition, all studies examining the use of modified HM in chylothorax included small samples (n ≤ 35), and it is possible that different groups may experience different results. Based on the available evidence, it appears that treatment of chylothorax with modified HM is a viable alternative to formula-based nutrition plans, with similar results related to chest tube volume and duration, treatment duration, LOS, and growth.

High-quality evidence including larger sample sizes is needed to clarify the impact of modified HM on outcomes related to chylothorax. It is also important to note that all four studies examining chylothorax outcomes in this review used a centrifuge to create low-fat HM,^28,36,37,39 with one of these providing parents with the option of portable centrifuges to use at home. ³⁹ It is understandable that many hospitals in North America many not have the resources to implement centrifuge-based skim milk programs for chylothorax. Institutions would require a strong HM culture, along with financial and personnel resources to implement this type of program. ⁴⁸

Other less expensive methods of modifying HM for chylothorax treatment have been described. ⁴⁹ Further testing of these methods in the CHD population is warranted, and should include qualitative methods to investigate any potential burden for families and hospital staff.

Weight gain

Impaired weight gain and physical growth can be a major concern for infants with CHD due to increased risk for poor surgical and developmental outcomes^4,5,50 and psychological stress for family caregivers.^6,7 It is difficult to compare growth outcomes between studies in this review, due to differences in feeding groups, outcome measures, and length of follow-up. For example, four studies compared “breastfeeding” or “breast milk” groups with “formula,” or “bottle” groups.^31,33,34,38 In three of these studies, infants in the breastfeeding groups were not required to receive exclusive HM.^31,33,34

Other studies examined HM feeding compared with standard or high-calorie formula, ³⁰ predominant breastfeeding compared with mixed feeding or exclusive formula, ³⁵ or HM-based fortifier compared with bovine-based fortifier. ⁴⁰ Older studies may not reflect current practices regarding postoperative or interstage fortification. Anthropomorphic measures also varied among studies (e.g., average daily weight gain,^30,31,38,40 percentage of weight change, ³⁴ and WAZ^33,35), and length of follow-up ranged from the initial postoperative hospital stay^30,38 to the first year of life. ³⁵

Despite these limitations in study design, there appears to be some evidence that a well-managed HM diet may support improved weight gain for infants with CHD. Blanco et al.'s recent RCT reported that infants with SV CHD who received HM-based fortifier had greater weight velocity from birth to the end of the study than those on bovine fortifier. ⁴⁰ Similar results have been found in other surgical congenital populations. ⁵¹ As Blanco et al.'s results were available only as a conference presentation, it was not clear whether the findings were adjusted to reflect the time that infants spent in the study, as length of participation varied due to infant clinical course. Full published results are needed to verify the strength of these findings.

Additional studies supported increased growth for infants receiving HM, with Yu et al. reporting significantly increased average weight gain for infants on exclusive HM diets, ³⁸ while Combs and Marino found that infants in HM feeding groups tended to lose less weight at 5 months postdischarge. ³⁴ Both of these studies exhibit risk for bias related to design and analysis. While McCrary et al. reported that infants fed any HM tended to lose less weight at shunt palliation takedown surgery (at mean 5.6 ± 2.7 months), the authors cautioned that weight gain was insufficient for all infants in their cohort, and baseline differences between the HM and non-HM groups were not well described.

The only study that demonstrated inferior weight-related outcomes for infants in HM feeding groups has serious design flaws (e.g., crossover between feeding groups; heterogeneous diagnoses and age groups; small sample; statistical analysis not well described) and does not reflect current practice. ³¹ The remaining studies revealed no significant differences in weight gain between infants receiving HM and those receiving other diets.^30,35 Taken as a whole, the results tentatively indicate that a well-managed HM diet—potentially including an HM-based fortifier when needed—may result in similar or improved weight-related outcomes for infants with CHD, although the evidence exhibits risk for bias.

High-quality, carefully designed studies are needed to fully understand the impact of HM feeding on weight gain and physical growth, so that best practices in managing an HM diet can be identified. Moreover, future research should investigate the impact of nutrition on the quality of growth outcomes for infants with CHD. Previous work in the preterm population reveals that infants fed with HM or direct breastfeeding experience slower initial weight gain, but ultimately better neurodevelopment ⁵² and recovery of body composition through fat-free mass deposition. ⁵³ Infants with CHD are at risk for impaired neurodevelopment, ⁵⁴ and inadequate growth has been shown to be associated with poor neurodevelopmental outcomes. ⁵⁰

However, the relationship between nutrition type (e.g., MOM, pasteurized donor HM, formula, and bovine or HM-based fortifier) and growth quality has not been examined in infants with CHD. The potential impact of HM feeding on neurodevelopment for these infants is an important direction for future research.

Other postoperative outcomes

The association between an HM diet and LOS was examined frequently, with five of six studies finding that infants in HM feeding groups had a shorter hospital LOS. The statistically significant result in Yu et al. should be interpreted with caution, as this study did not control for potential covariates (e.g., age and diagnosis). ³⁸ In addition, the authors note that infants in the formula feeding group often lived in rural areas, with parents who could not always travel to the surgical center. It is possible that this disparity in health care access could have delayed discharge or may have impacted outcomes in other unmeasured ways.

Future studies with large sample sizes and robust analytical techniques should carefully examine the impact of HM on hospital LOS, as a reduction in LOS by even 1 day has clinical implications for lower hospital costs and benefit to families.

Yu et al. also reported differences in other postoperative feeding outcomes, with infants in the HM groups beginning enteral feeds earlier and achieving full feeds sooner. ³⁸ Similar results in faster achievement of full feeds have been noted in some preterm populations.^55,56 For Yu et al., criteria for beginning enteral feeds included resumption of bowel sounds, typically on postoperative day 2–4, with feeds advanced based on gastric residuals. The authors present hypotheses for faster achievement of full feeds, but do not discuss potential reasons for earlier feeding readiness in the HM group, which could have been impacted by unmeasured covariates (e.g., variation in feeding based on individual clinician preference and level of family engagement/advocacy for feeding).

Multisite studies with larger samples are needed to corroborate these findings, particularly as postoperative feeding practices are highly site and provider specific. ⁹ Yu et al. found additional benefits of HM feeding related to markers of nutritional status and incidence of complication. ³⁸ Many of these analyses included low numbers and were likely not powered to detect true differences. Further evidence is needed to validate the results described in this study.

Definition of HM feeding

Studies focused on HM feeding have been historically imprecise in defining what constitutes an HM feeding group. ⁵⁷ The 16 studies in this review exhibit notable variation in delineating HM feeding, with 10 studies providing either no clear definition of HM feeding,^19,35,37,41 or allowing infants in HM groups to consume significant amounts of formula.^{28,31,33,34,36,39} While it should be noted that examining HM feeding outcomes was not the primary aim of Becker et al., this study assessed infant nutrition type on the day of NEC diagnosis, and assigned 5 (24%) infants to HM or formula feeding groups based on this data point. ³² The remaining 16 (76%) infants diagnosed with NEC were not included in the analysis due to missing feeding data on the day of diagnosis.

The ambiguity in feeding group assignment in the studies under review complicates pooling of results and cross-study comparison. While there are ethical challenges inherent in researching HM feeding, it is difficult to make recommendations based on studies with multiple treatment interference, or with large amounts of missing data. Several recent studies addressed this problem by providing pasteurized donor HM for infants who were not able to receive 100% MOM,^27,29,40 although it should be noted that pasteurized donor HM is not an equivalent substitute for MOM.^58–60

Future studies should explore HM dose/response. Many studies in the current review allowed infants in an HM group to consume significant amounts of formula (e.g., range of up to 65% in Kocel et al. ³⁶ ), or considered an HM feeding group to include any amount of HM.^31–34 Kataria-Hale et al. hypothesize that their nonsignificant association between preoperative HM feeding and postoperative NEC may have been different if they had been able to calculate HM dose. ²⁹

To our knowledge, no study has quantified the amount of HM received by infants with CHD or analyzed volume-associated differences in outcomes. Accurate methods to measure HM transferred via direct breastfeeding exist (e.g., test weighing), ⁶¹ and can be used to quantify volume for those infants who are not receiving exclusive expressed HM.

Harms

In the included studies, there were no harms associated with an HM diet. There is a need for well-designed, prospective research on the best ways to support long-term physical growth through an HM diet in this population, but the results in this group of studies indicate that an HM diet is not harmful to growth or development, and may in fact support physical growth. Given the well-documented nutritional, immunological, and relational benefits of an HM diet in both the healthy newborn population and in those needing NICU care, ⁴³ the benefits of a well-managed HM diet outweigh any potential risks.

Clinical implications

Considering the evidence supporting decreased odds for preoperative NEC in infants with CHD who receive exclusive HM feeding, the possibility that an HM diet may decrease hospital LOS, the potential for improved postoperative feeding and nutrition outcomes, and the finding that a well-managed HM diet may support improved growth, we recommend that clinicians and health care systems make a concerted effort to support an exclusive HM diet, potentially fortified with HM-based fortifier when needed, for infants with CHD both during the hospital stay and after discharge. This aligns with global health organizations, which agree that HM is the optimal nutrition for both typically developing and vulnerable infants.^15,16,62

Clinicians should identify ways to support lactating parents in providing HM for their infant with CHD. Direct breastfeeding and HM expression are inherently challenging in the context of neonatal surgery and hospitalization,^11,63 and infants with CHD are at particular risk for early weaning from HM.^17,64 Furthermore, infants with CHD are often cared for at children's hospitals, potentially in a cardiovascular ICU. These institutional environments may not have a strong culture of prioritizing HM feeding and direct breastfeeding, in comparison with a birth hospital or an NICU. Previous research has demonstrated that parents of vulnerable infants born in a specialized birthing center inside a children's hospital were significantly more likely to initiate pumping and HM feeding for their infant with CHD, compared with infants born at an outside hospital. ⁶⁵

This systematic review could serve as a call to action for institutions caring for infants with CHD. First, health care providers in these facilities should examine their current lactation services and ensure that they are providing comprehensive, evidence-based lactation interventions during the antenatal period, through birth, during the infant's hospitalization, and postdischarge. Second, health care institutions should examine their process of documenting enteral nutrition. Ideally, every enteral feed should be quantified to determine percentages of MOM, pasteurized donor HM, and infant formula so that percent exposure to HM could be calculated daily, weekly, and over the entire hospital stay.

Third, hospitals caring for infants with CHD should consider investing in the necessary equipment and personnel to implement a skim milk program. For an infant with CHD, the early weeks of life are often a time of crucial interventions, intended to optimize growth and development throughout the life span. Support of HM feeding should be a key component of this early development plan, requiring a concerted interdisciplinary, family-centered approach, including comprehensive, culturally responsive lactation services that include prenatal intervention, support for parental mental health, and evidence-based care. ⁶⁶

Directions for future research

The current review demonstrates that there are major gaps in the literature to be addressed through future research. The overall quality of the current evidence has limitations, with issues of statistical power and analysis affecting the group of studies.

In addition, studies used varying or unclear criteria for feeding group assignment, with infants in HM feeding groups often receiving significant amounts of non-HM nutrition for supplementation or fortification. While fortification may be necessary for some infants with CHD, future research should be clear about the percentage of non-HM feeding that is acceptable for an infant categorized in an HM study group. Ideally, improvements in enteral nutrition measurement and documentation would allow for studies to determine whether there are dose/response differences in outcomes between infants receiving HM, formula feeding, and/or bovine-based fortifier.

While some recent studies exhibit more clarity in HM feeding group delineation,^27,39,40 this review highlights the fact that CHD is inherently difficult to study. Physiological combinations of anomalies may be highly individualized, and two patients with the same diagnosis may experience very different clinical courses. ⁶⁷ Research on nutrition for hospitalized infants with CHD is similarly complicated, as feeding plans can include various forms of nutrition and routes of nutrition delivery. It is challenging to tease apart the factors that interact to impact each infant's experience with CHD. Finally, clear standardization of outcome measurements would improve the quality of the evidence, particularly for rare diagnoses such as NEC and for growth-related outcomes.

Large multisite studies implementing high-quality, adequately powered prospective designs are needed to fully understand the impact of an HM diet on outcomes for infants with CHD, with robust statistical analysis that carefully controls for clinically relevant covariates. These studies should aim to clarify the current inconclusive evidence, with a particular focus on outcomes related to growth and physical development. Growth failure is associated with increased morbidity and mortality in infants with CHD,^4,5 and a recent study highlighted the lack of evidence for best practices in managing growth with an HM diet in this population. ²¹

Therefore, future research should focus on exploring techniques for managing an infant's HM diet, which may include individualized fortification or separating the hindmilk of MOM.^68,69 Future work should focus on describing the longitudinal impact of an HM diet on physical growth and development, and should consider dose-dependent outcomes. Multisite studies can provide power to determine differences in rare outcomes such as NEC; however, standardization of outcome measurements among sites is a key consideration, and may be challenging to implement.

Of note, fewer than half of studies in this review reported race or ethnicity,^{19,27–29,32,33,35} and only three offered any type of information on social or economic characteristics of the sample.^33,35,36 There may be unexplored social determinants of health (e.g., structural racism and structural inequity) or barriers to health care access (e.g., distance from center and language barriers) that could play a role in the relationship between infant nutrition and health outcomes. ⁷⁰ For example, Yu et al. mention that many of the infants in their formula feeding group came from rural families, who lived far from the tertiary hospital and could not travel to provide HM for their child. ³⁸ This raises questions about differences in parental proximity or engagement that may have impacted health outcomes.

Future studies should report social and familial information, and should interrogate the ways in which systems and institutions contribute to inequities in HM-associated outcomes for infants with CHD, particularly given that factors such as parental race, country of birth, level of education, neighborhood, socioeconomic status, and maternal parity have been shown to impact HM feeding and direct breastfeeding in other vulnerable neonatal populations.^71–73

In addition, the impact of parental mental health on infant feeding and health outcomes cannot be ignored. Previous literature demonstrates the negative impact of emotional and physiological stress on HM production,^74,75 with higher levels of parental engagement and lower levels of anxiety and depression associated with improved HM outcomes. ⁷⁶ Considering the potential for parental trauma in the context of a neonatal hospitalization, future work should adopt a holistic, culturally responsive, family-centered approach to HM research that acknowledges infant feeding as an innate, nurturing parental act of care, as opposed to merely a medical intervention to be measured and documented. ⁷⁷

Strengths and limitations

The current review has several strengths. The search was conducted in multiple databases, with the assistance of an experienced health sciences research librarian. Reference lists were carefully examined for relevant publications. While non-English-language studies were excluded, a title review of non-English results identified no studies that appeared to meet the inclusion criteria. The review process included an in-depth methodological quality assessment by at least two independent reviewers, following the guidelines of validated critical appraisal tools. ²⁶

Limitations include risk for bias in many studies; only one non-pilot RCT; a lack of information on social determinants of health and health inequities; and inconsistency in defining HM feeding among studies. The precise type of nutrition (e.g., MOM or donor HM) was often unclear. Sample sizes varied and many studies were not well powered, with subsequent issues for statistical validity.

While the review represents diverse CHD diagnoses, the exclusion criteria of several studies raise questions about the generalizability of the findings. For example, Yu et al. ³⁸ report several significant results, yet include no infants with staged surgery for CHD, effectively eliminating infants who could be at the greatest risk for feeding-related morbidity and mortality. Similarly, DiLauro et al. ³⁹ decided not to approach some unstable HM-fed infants undergoing treatment for chylothorax. Therefore, the results of this review may not fully describe the wide range of clinical experiences in the CHD population, with the sickest infants possibly underrepresented.